Pharmpy in R

pharmr is an R wrapper to Pharmpy. It provides an R interface to all functions found in the modeling-module and the tools-module. Each function is also available via the help-function (or ?). pharmr is available on CRAN, however, since Pharmpy is under rapid development, we still recommend using the development version on Github until Pharmpy/pharmr version 1.0.0 has been released.

Installing pharmr

Install pharmr and Pharmpy with the following:

remotes::install_github("pharmpy/pharmr", ref="main")
pharmr::install_pharmpy()

It is also possible to install from CRAN (but we still recommend using the GitHub installation until 1.0.0 is out):

install.packages("pharmr")
pharmr::install_pharmpy()

Note

Make sure to clear your environment and restart the R session after installing/updating.

To make sure the versions of Pharmpy and pharmr are up-to-date and were installed successfully, run the following:

library(pharmr)
packageVersion("pharmr")
print_pharmpy_version()

If you get any error during the installation or when running the code above, please check Trouble shooting.

Trouble shooting

Wrong Python version when using conda

pharmr uses the package reticulate for calling Python from R. When reticulate sets up Miniconda it can default to use Python 3.8. If you have any trouble installing Pharmpy or any of its dependencies, you can do the following to check the Python version in your reticulate environment:

reticulate::py_discover_config()

Make sure the Python version is >= 3.10. If it is not, you can run the following in R:

reticulate::conda_create('r-reticulate', python_version = '3.11')

Restart the session and try installing Pharmpy again:

pharmr::install_pharmpy()

Error importing Pharmpy

If you have followed the installation steps but still get an error asking you to install Pharmpy, the issue might be that pharmr is not using or finding the right virtual environment. If you used pharmr to set up your environment (default when you use install_pharmpy()), run the following:

reticulate::py_discover_config()

and make sure ‘r-reticulate’ is found:

python:         .../r-reticulate/bin/python
libpython:      .../r-reticulate/lib/libpython3.11.so
...

If you are using Rstudio you can change this in either project or global options, under Python/Python interpreter.

Mismatch of versions between Pharmpy and pharmr

The version number of pharmr mirrors Pharmpy, so it is important to make sure they have the same version number (a warning will appear if they are different). To avoid this happening, use install_pharmpy() which detects which version of pharmr you have installed and installs the correct version.

Using pharmr

In pharmr, you can pipe different Pharmpy functions together with the magrittr-package:

library(pharmr)
library(magrittr)
model <- read_model('path/to/model') %>%
  set_zero_order_absorption() %>%
  fit()

Gotchas

List indices

One difference between Python and R is that in Python, list indices start at 0 while in R it starts at 1. Since Pharmpy is developed in Python, in functions where you have arguments referring to indices, you need to use the Python way. For example:

set_estimation_step(model, method, interaction = TRUE, options = NULL, idx = 0)

Similarly, you need to use 0-indexing when accessing the first element of Pharmpy objects such as random variables:

rvs <- model$random_variables
rvs[0] # access first element